Contribution of archaeological analogues to the comprehension of long term corrosion of concrete reinforcements

The study of archaeological analogues is necessary to improve the knowledge on the long-term corrosion of low carbon steels that could be used in concrete to build the structures of nuclear waste storage facilities. The long-term corrosion system was previously described as a multi-layer pattern made of the Metal, the Dense Product Layer (constituted of goethite with magnetite and/or maghemite marblings -DPL), the Transformed Medium (TM) which is an interphase between the Dense Product Layer, and the last layer, the Binder. As mainly constituted of goethite, a non-conductive phase, and assuming that the DPL pores are saturated with water, corrosion kinetic could be limited by oxygen diffusion in the water of the pores. Moreover, anodic and cathodic reactions should occur at the Metal/DPL interface. This last hypothesis was verified labelling oxygen precipitation sites with an oxygen isotope ( 18 O) and observing an increase of oxygen isotopic ratio at the M/DPL interface. Then, to validate the hypothesis of a diffusion control of corrosion kinetic, some investigations were performed on diffusion properties of oxygen in the DPL. For this purpose, the oxygen apparent diffusion coefficient was evaluated using diffusion cells made with DPL sampled on archaeological analogues. Then, thanks to Faraday law, it was possible to evaluate instantaneous corrosion rates (less than 0.1 μm/year). The compatibility of these rates with those obtained considering the quantity of 18 O precipitated in corrosion products was discussed. Nevertheless, even if oxygen diffusion is probably the limiting factor on iron corrosion in the very specific case of water saturated media, the hydrometry of the DPL during the life time of the object and the influence of this hydrometry on corrosion mechanisms have to be verified. Moreover, several questions were raised by TM formation and growth. Was this layer formed dissolving/precipitating corrosion products from the DPL? Unfortunately, thermodynamical data and results from our experimentation show that iron quantity obtained by this mean is not sufficient to explain TM thicknesses observed and, so far, TM formation mechanisms are still to be identified.